Iran's comprehensive heat flow map generated by the Random Forest method and the Sequential Gaussian Simulation

Abstract

The objective of this research is to present a comprehensive heat flow map for Iran by utilizing a novel approach that combines geostatistical distributions, machine learning algorithms, and deterministic calculations. In order to enhance the accuracy and quality of the initial geothermal map obtained from a recent study, a machine learning technique called the Random Forest method was employed in the mapping process, taking into consideration approximately 100 processed data points. To accomplish this, various geological features such as active faults, hot springs, geological domains, and hot spots were identified as effective attributes for generating a geology-based geothermal map. This geology-based map was then used as a trend to create a geostatistics-based geothermal gradient map. The sequential Gaussian co-simulation method was adopted for the geostatistical distribution, enabling the extension of the available geothermal gradient data points (which were calculated in a recent study) to cover the entire surface of Iran on a map. Consequently, an integrated geothermal gradient map was derived. Afterwards, thermal conductivity values corresponding to the prevalent lithology of each location in Iran were incorporated to calculate the heat flow values. This process resulted in the generation of a comprehensive heat flow map. The introduction of such an extensive map would greatly contribute to future geothermal-related explorations and complementing research in this field.